Web sites are (most often) not refereed journals or conferences. This means that the document referenced could be full of incorrect and/or misleading information. A good journal or conference is the proper place to reference (and publish) scientific information. If the article appears in a journal and the web, one should reference the journal.

But as the web grows so does the fact that the only place many documents are available is on the web. So what is the proper (or good) way to reference web sites? The basic problem is that a web version of a document is not always permanent. Libraries are generally more reliable in this respect. Enough information must be given in the reference to allow researchers access to the correct document. Below is an example:

Maletic, J., "Some Notes on Citing Web References", Nov. 1998, Available [online]: http://www.cs.kent.edu/~jmaletic/webrefs.html, Date accessed: 9/22/1999.

This gives most of the important information. You'll need the author(s) of the document, when it was FIRST published, and where to get it. "Available [online]:" is a good keyword descriptor for this purpose, and could support other locations (newgroups, listservers, etc.). The date accessed is VERY important. This gives a "version number" of the document to which you are referring. For a more discussion and description see:

http://www.apastyle.org/apa-style-help.aspx